Vehicle information identification method and device based on artificial intelligence anti-interference

Abstract

The invention discloses a vehicle information identification method and device based on artificial intelligence anti-interference, relates to the technical field of artificial intelligence, and mainly aims to extract the identification accuracy of a vehicle. The method comprises the following steps: extracting a candidate area of a suspected vehicle from a video monitoring image based on a preset extraction standard; if the candidate area is not extracted, inputting the video monitoring image into a pre-trained artificial intelligence recognition model to recognize whether the candidate area exists in the video monitoring image or not, wherein the artificial intelligence recognition model is trained according to a sample vehicle image library carrying interference information and a neural network model; if the candidate region exists in the video monitoring image, performing adaptive adjustment on a preset extraction standard, and extracting the candidate region from the video monitoring image for the second time based on the adjusted extraction standard; and identifying whether the candidate area obtained by the secondary extraction is a vehicle or not by utilizing a preset vehicle identification template. The method is suitable for vehicle information identification based on artificial intelligence anti-interference.

Description

technical field [0001] The invention relates to the technical field of city monitoring based on artificial intelligence, in particular to a vehicle information identification method and device based on artificial intelligence anti-interference. Background technique [0002] In recent years, with the continuous improvement of people's requirements for safety or quality of life, smart cities have begun to appear. In many aspects of smart cities, such as traffic monitoring, security prevention and control, community services, etc., it is necessary to identify or extract information from video surveillance screens. Vehicle Information. At present, usually according to a fixed preset extraction standard, a candidate area that may be a vehicle is extracted from a video surveillance screen, and then whether the candidate area belongs to a vehicle is identified according to whether the extracted candidate area matches a predetermined template, and Further identify the vehicle type ...

AI Technical Summary

Problems solved by technology

[0004] However, in practice, there will be some intentional or unintentional interference information on the car body. The features are used as interference information, and these interference information will cause the candidate area corresponding to the vehicle to be inconsistent with the preset extraction standard, and cause the area that is essentially a vehicle to not be recognized as a candidate area, so that the vehicle identification is missed, resulting in incorrect vehicle identification results. Accurate, which leads to low recognition accuracy of the vehicle

Images